Project Summary The long-term objective of this project is to develop a dynamical model of visually controlled locomotor be- havior, which can be used to simulate visual mobility problems and predict paths of locomotion in everyday environments. The research pursues two closely related problems: First, the behavioral dynamics of locomotion, or how safe paths of locomotion are generated through a complex changing environment; second, the control laws for locomotion, or how this behavior is guided by perceptual control strategies based on multisensory in- formation. In previous research, we developed a dynamical model of four elementary locomotor behaviors: (a) steering to a stationary goal, (b) avoiding a stationary obstacle, (c) intercepting a moving target, and (d) avoiding a moving obstacle. Using virtual reality techniques in freely-walking participants, we empirically tested each component, and then studied how the elementary behaviors are integrated in more complex situations. We found that linear combinations of the components successfully predict human paths with various pairs of goals and obstacles, such as a moving target with a stationary obstacle, and even with pairs of participants in a pursuit/evasion task; however, attentional factors appear to play a role with multiple obstacles. In a second set of studies, we tested the contributions of optic flow, egocentric direction, and gaze/head/body alignment to online steering control and visual-locomotor adaptation. The specific aims of the proposed research are to pursue locomotor interactions between pedestrians, map out multisensory contributions to perceptual control strategies, and apply the model to simulate visual im- pairments and mobility problems. We will manipulate visual information during walking using virtual envi- ronments presented in a head-mounted display, podokinetic information by locomotor adaptation, vestibular information by galvanic stimulation, and neck proprioception by muscle vibration. Four studies are proposed: (1) Pedestrian interactions will investigate locomotor interactions between pedestrians, including a pair of par- ticipants, a participant interacting with a simulated virtual agent, a participant interacting with a crowd of vir- tual agents, and naturalistic studies of pedestrian traffic flow. (2) Multisensory control strategies will test the contributions of visual, podokinetic, vestibular, and proprioceptive information to steering control. (3) Com- plex behaviors will extend the model to incorporate attentional factors, speed control, and address the question of online control or advance path planning. (4) Modeling peripheral field loss will empirically test and seek to model the effects of tunnel vision and homonymous hemianopsia on mobility. 14